What Research Says About Online Learning

Share This:

Facebook

Twitter

LinkedIn

Email

September 12, 2025

Summary: Online learning has become a core component of education at all levels. The COVID-19 pandemic created a massive natural experiment, offering researchers more data about how students react to content. That research offers several insights into what works in online education. Artificial intelligence is increasingly capable of creating personalized lesson plans for students.

Over the past couple of decades, online learning has moved from a niche alternative to a core component of education. That shift accelerated drastically during the pandemic, which affected over 1.6 billion learners worldwide, hitting disadvantaged students hardest.

Researchers rushed to share ideas on how best to meet the needs of students around the world. To get a perspective, consider this: the journal IEEE Transactions on Education saw manuscript submissions increase by 34% from 2021 to 2022, and expanded the number of issues it published each year. 

While online learning may not have been the only factor that impacted an increase in manuscripts and issues published, the global body of research helped to provide a new understanding of what works in online education, based on detailed analyses of how students interacted with course materials.

“Online learning works best when it feels less like sitting through a lecture and more like playing a game,” said IEEE Senior Member Shaila Rana. “The most exciting part here is that it isn’t just better tech, it’s rethinking what school and education could be.”

It’s clear that it is here to stay and growing. In 2022, as college campuses in the U.S. reopened, more than half of students reported taking at least one online class. Businesses rely on microcredentials and short learning systems to improve the skillset of their workforce. 

From AI-powered personalization to immersive virtual laboratories, the transformation goes far beyond simply moving classrooms online.

Factors for Student Success

Online learning platforms have access to a lot of data, including forum posts that students write and keystrokes they make when they type. The most important source of data, however, has historically been the clickstream, a meticulous log of pages visited, the order in which students visited them, the time spent on specific resources and where students stopped and started videos and assignments. 

A report in IEEE Access, Innovations in Online Learning Analytics: A Review of Recent Research and Emerging Trends, documents the history of this analytical stream. It notes that over time, online course development has evolved from pinpointing the specific material that caused students to struggle to identifying students at risk of failing or dropping out using predictive analytics. 

That, and other research, also notes that: 

• Artificial intelligence is increasingly capable of creating personalized learning plans, tailoring lessons to students’ progress. 

• In large courses with thousands of students, natural language processing techniques can identify sources of student frustration. 

• A faster feedback cycle is key to keeping students engaged. One emerging area is the use of automated essay scoring to accelerate a first-person review, with a more in-depth review coming later.

• By combining eye-tracking data and the clickstream, course designers can evaluate student boredom and engagement. The result is the emergence of hybrid learning models, which blend short, in-person sessions with self-paced online modules. 

• Proper data management is crucial. Analysts need to make sure the data sets they use to evaluate online courses draw from a broad range of students, not just high performers. 

Personalized Learning with AI 

In traditional education methods, every student is taught the same material on the same schedule. In the early days of online learning, instruction followed a similar pattern, with an emphasis on Massive Open Online Courses. These classes generally focused on foundational topics that hundreds or thousands of students would be required to take in a particular field, such as “Introduction to Computer Programming.”  

Technology has opened the door for online education to take the opposite direction, with more personalization and coursework structured to students’ needs. 

AI-based personalized learning systems for e-learning provide students with individual learning methods by assessing a learner’s level and comprehension and determining appropriate content for success. This is specifically useful where online education supplements physical classes. In addition, personalized e-learning systems can also be implemented to educate the masses, as they provide a cost-effective method of delivering education. 

Rana has seen this transformation happening in real time.

“Smart algorithms can watch how students learn and automatically serve up easier or harder problems based on their performance,” she said. “AI can spot patterns in student behavior that humans miss — such as when someone always struggles with math problems on Friday afternoons or stops participating before big deadlines.”

Did you know? Over the last six years, more than 700,000 learners have enrolled in courses through the IEEE Learning Network. They’ve taken over 1,800 classes, with English for Technical Professionals ranking as the most popular. From 2024 to 2025, there’s been a 116 percent increase in online learners.